Abstract: An embodiment of the present invention includes an adaptive predictor, a system white noise generator, and a background noise estimator. The adaptive predictor estimates adaptive weights of autoregressive (AR) model of background noise as background samples in an echo canceler. The adaptive predictor generates adaptive error. The system white noise generator generates a white noise using the adaptive error. The noise background estimator estimates the background noise using the white noise and the estimated adaptive weights.

Abstract: A method and apparatus for adapting dual filters is disclosed. In one aspect, a method may include transforming a signal, adapting a first adaptive filter based on the transformed signal, estimating a delay of an impulse response based on the adaptation of the first filter, delaying a signal based on the estimated delay, and adapting a second adaptive filter based on the delayed signal. In one aspect, an echo or other unwanted signal may be reduced or cancelled based on the adaptation of the second filter.

Abstract: The present invention is an echo canceller for sparse channels in an echo channel. The echo channel receives a far-end input and a near-end input. An enable generator asserts a sparse update enable in response to detecting at least one of the double talk condition and an impulse response change. A main weight updater transfers shadow weights of the shadow filter to N sets of sparse weights in N sparse filters of the main filter according to the sparse update enable. A sparse main filter filters the far-end input non-adaptively to generate a main output. A long shadow filter filters the far-end input adaptively to generate a shadow output and to estimate delays of peaks of the sparse weights in the main filter.

Abstract: The present invention is an echo canceller for sparse channels in an echo channel. The echo channel receives a far-end input and a near-end input. An enable generator asserts a sparse update enable in response to detecting at least one of the double talk condition and an impulse response change. A main weight updater transfers shadow weights of the shadow filter to N sets of sparse weights in N sparse filters of the main filter according to the sparse update enable. A sparse main filter filters the far-end input non-adaptively to generate a main output. A long shadow filter filters the far-end input adaptively to generate a shadow output and to estimate delays of peaks of the sparse weights in the main filter.

Abstract: An embodiment of the present invention includes an adaptive predictor, a system white noise generator, and a background noise estimator. The adaptive predictor estimates adaptive weights of autoregressive (AR) model of background noise as background samples in an echo canceler. The adaptive predictor generates adaptive error. The system white noise generator generates a white noise using the adaptive error. The noise background estimator estimates the background noise using the white noise and the estimated adaptive weights.

Abstract: The invention provides a novel scheme for performing echo cancellation in the presence of double-talk and near-end channel impulse response changes. In one embodiment of the invention, a non-adaptive main filter is updated with the filtering weights of an adaptive shadow filter if the shadow filter cancels near-end echo in a first signal to a greater extent than the main filter. However, if double-talk is present in the first signal, then the non-adaptive filter is not updated. According to one embodiment of the invention, distinguishing between double-talk and channel impulse response changes is accomplished by maintaining extra taps for the main and shadow filters, in addition to taps employed by the main and shadow filters for echo canceling. The corresponding filtering algorithm weights for the additional taps of the main and shadow filters are compared to detect the onset of double-talk and/or channel impulse response changes.